import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt

# Esses dados de churn ja oferece dois datasets um de 'treino' e um de 'teste'
# Vamos usar o de treino para adquirir todos os insights e depois usar os dois na parte de ML
# Tem outro dataset, o sampleSubmission, como peguei esses dados de uma competição no Kaggle, o sampleSubmission é como se fosse um 'gabarito
# De forma resumida vamos fazer tudo com o treino, depois vamos usar o test para simular uma nova predição, e depois usar...
# o gabarito para ver o score...podemos fazer tudo com treino e depois so usar o modelo no teste tambem.

train = pd.read_csv('dataset/train.csv')
test = pd.read_csv('dataset/test.csv')

display(train.head(3), train.shape)

	state	account_length	area_code	international_plan	voice_mail_plan	number_vmail_messages	total_day_minutes	total_day_calls	total_day_charge	total_eve_minutes	total_eve_calls	total_eve_charge	total_night_minutes	total_night_calls	total_night_charge	total_intl_minutes	total_intl_calls	total_intl_charge	number_customer_service_calls	churn
0	OH	107	area_code_415	no	yes	26	161.6	123	27.47	195.5	103	16.62	254.4	103	11.45	13.7	3	3.70	1	no
1	NJ	137	area_code_415	no	no	0	243.4	114	41.38	121.2	110	10.30	162.6	104	7.32	12.2	5	3.29	0	no
2	OH	84	area_code_408	yes	no	0	299.4	71	50.90	61.9	88	5.26	196.9	89	8.86	6.6	7	1.78	2	no

(4250, 20)

# Eu poderia dar join (merge no pandas) para ficar em so um...mas vou deixar assim

sampleS = pd.read_csv('dataset\sampleSubmission.csv')

display(test.shape, sampleS.shape)

display(test.head(1), sampleS.head(1))

(750, 20)

(750, 2)

	id	state	account_length	area_code	international_plan	voice_mail_plan	number_vmail_messages	total_day_minutes	total_day_calls	total_day_charge	total_eve_minutes	total_eve_calls	total_eve_charge	total_night_minutes	total_night_calls	total_night_charge	total_intl_minutes	total_intl_calls	total_intl_charge	number_customer_service_calls
0	1	KS	128	area_code_415	no	yes	25	265.1	110	45.07	197.4	99	16.78	244.7	91	11.01	10.0	3	2.7	1

	id	churn
0	1	yes

1.1 EDA

train.describe()

	account_length	number_vmail_messages	total_day_minutes	total_day_calls	total_day_charge	total_eve_minutes	total_eve_calls	total_eve_charge	total_night_minutes	total_night_calls	total_night_charge	total_intl_minutes	total_intl_calls	total_intl_charge	number_customer_service_calls
count	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000	4250.000000
mean	100.236235	7.631765	180.259600	99.907294	30.644682	200.173906	100.176471	17.015012	200.527882	99.839529	9.023892	10.256071	4.426353	2.769654	1.559059
std	39.698401	13.439882	54.012373	19.850817	9.182096	50.249518	19.908591	4.271212	50.353548	20.093220	2.265922	2.760102	2.463069	0.745204	1.311434
min	1.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
25%	73.000000	0.000000	143.325000	87.000000	24.365000	165.925000	87.000000	14.102500	167.225000	86.000000	7.522500	8.500000	3.000000	2.300000	1.000000
50%	100.000000	0.000000	180.450000	100.000000	30.680000	200.700000	100.000000	17.060000	200.450000	100.000000	9.020000	10.300000	4.000000	2.780000	1.000000
75%	127.000000	16.000000	216.200000	113.000000	36.750000	233.775000	114.000000	19.867500	234.700000	113.000000	10.560000	12.000000	6.000000	3.240000	2.000000
max	243.000000	52.000000	351.500000	165.000000	59.760000	359.300000	170.000000	30.540000	395.000000	175.000000	17.770000	20.000000	20.000000	5.400000	9.000000

train.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 4250 entries, 0 to 4249
Data columns (total 20 columns):
 #   Column                         Non-Null Count  Dtype  
---  ------                         --------------  -----  
 0   state                          4250 non-null   object 
 1   account_length                 4250 non-null   int64  
 2   area_code                      4250 non-null   object 
 3   international_plan             4250 non-null   object 
 4   voice_mail_plan                4250 non-null   object 
 5   number_vmail_messages          4250 non-null   int64  
 6   total_day_minutes              4250 non-null   float64
 7   total_day_calls                4250 non-null   int64  
 8   total_day_charge               4250 non-null   float64
 9   total_eve_minutes              4250 non-null   float64
 10  total_eve_calls                4250 non-null   int64  
 11  total_eve_charge               4250 non-null   float64
 12  total_night_minutes            4250 non-null   float64
 13  total_night_calls              4250 non-null   int64  
 14  total_night_charge             4250 non-null   float64
 15  total_intl_minutes             4250 non-null   float64
 16  total_intl_calls               4250 non-null   int64  
 17  total_intl_charge              4250 non-null   float64
 18  number_customer_service_calls  4250 non-null   int64  
 19  churn                          4250 non-null   object 
dtypes: float64(8), int64(7), object(5)
memory usage: 664.2+ KB

# Temos um dataset meio desbalanceado, tem muito mais 'Não CHURN' do que 'CHURN'.
display(train['churn'].value_counts())

plt.pie(x = train['churn'].value_counts(), labels = ['Not Churned', 'Churned'], autopct='%.2f')
plt.legend()
plt.show()

no     3652
yes     598
Name: churn, dtype: int64

# Na verdade eu vou aproveitar para ver todas as variaveis categoricas, so pra ter um resumo delas e decidir oq fazer
# Menos os estados...não vejo muito sentido em olhar eles
# Acabei colocando o churn novamente, mas é bom que fica bonito o grafico

fig, axs = plt.subplots(2, 2, figsize = (16,9))

sns.countplot(x = train['area_code'], data = train, ax = axs[0,0])
axs[0,0].bar_label(axs[0,0].containers[0], padding = -15)

sns.countplot(x = train['international_plan'], data = train, ax = axs[0,1])
axs[0,1].bar_label(axs[0,1].containers[0], padding = -15)

sns.countplot(x = train['voice_mail_plan'], data = train, ax = axs[1,0])
axs[1,0].bar_label(axs[1,0].containers[0], padding = -15)

sns.countplot(x = train['churn'], data = train, ax = axs[1,1])
axs[1,1].bar_label(axs[1,1].containers[0], padding = -15)

[Text(0, -15, '3652'), Text(0, -15, '598')]

# Dar uma olhada como as variaves se correlacionam
# Tem umas variaveis que sao categoricas

corr =  train.corr().round(3)

mask = np.triu(np.ones_like(corr, dtype=bool))

f, ax = plt.subplots(figsize=(16, len(train.columns)))
sns.heatmap(data = corr, mask = mask,annot=True,cmap='coolwarm', vmin=-1, vmax=1, center = 0, square = True, linewidths=0.6, cbar_kws={"shrink": .5})
ax.set_title('Matriz de correlação', fontsize = 25)
plt.show()

# Ja vou aproveitar pra tirar os features que tão com corr = 1 !
# Isso acontece provavelmente pois o charge esta diretamente ligado ao numero de minutos que a pessoa passa
# Escolhi deixar o charge, não sei pq vou ser sincero...

train = train.drop(['total_day_minutes','total_eve_minutes','total_night_minutes','total_intl_minutes'], axis=1)
test = test.drop(['total_day_minutes','total_eve_minutes','total_night_minutes','total_intl_minutes'], axis=1)

sns.pairplot(train, vars=['account_length','number_vmail_messages', 'total_day_charge',
                          'total_eve_charge','total_night_charge','total_intl_charge','number_customer_service_calls'],hue='churn')

<seaborn.axisgrid.PairGrid at 0x1791e95b280>

1.2 Normal Distributions

# Eu ia checar um por um pra ver se tem algum feature que precisa tratamento, mas vou plotar mesmo
from scipy.stats import shapiro
shapiro(train['total_intl_charge'])

ShapiroResult(statistic=0.9930943250656128, pvalue=1.831744391583845e-13)

# Por um momento eu pensei em fazer um por um...mas ai a eu pensei melhor e fiz um for mesmo
# sns.displot(train['account_length'], kde = True)
# plt.show()

for feature in train:
    if train[feature].dtype != ('O'):
        plt.figure(figsize = (4,4))
        sns.displot(train[feature], kde = True)
        plt.title(feature)
        plt.show()

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

<Figure size 400x400 with 0 Axes>

1.3 Outlier Detection

for feature in train:
    if train[feature].dtype != ('O'):
        plt.figure(figsize = (4,4))
        sns.boxplot(x ='churn', y = feature, data = train)
        plt.title(feature)
        plt.show()

# https://careerfoundry.com/en/blog/data-analytics/how-to-find-outliers/
def find_outliers_IQR(df):
    q1=df.quantile(0.25)
    q3=df.quantile(0.75)
    IQR=q3-q1
    outliers = df[((df<(q1-1.5*IQR)) | (df>(q3+1.5*IQR)))]
    return outliers

def drop_outliers_IQR(df):
   q1=df.quantile(0.25)
   q3=df.quantile(0.75)
   IQR=q3-q1
   not_outliers = df[~((df<(q1-1.5*IQR)) | (df>(q3+1.5*IQR)))]
   outliers_dropped = not_outliers.fillna(not_outliers.mean()).reset_index()
   return outliers_dropped

train = drop_outliers_IQR(train)

C:\Users\Caio1\AppData\Local\Temp\ipykernel_11216\4227797413.py:13: FutureWarning: Automatic reindexing on DataFrame vs Series comparisons is deprecated and will raise ValueError in a future version. Do `left, right = left.align(right, axis=1, copy=False)` before e.g. `left == right`
  not_outliers = df[~((df<(q1-1.5*IQR)) | (df>(q3+1.5*IQR)))]
C:\Users\Caio1\AppData\Local\Temp\ipykernel_11216\4227797413.py:14: FutureWarning: Dropping of nuisance columns in DataFrame reductions (with 'numeric_only=None') is deprecated; in a future version this will raise TypeError.  Select only valid columns before calling the reduction.
  outliers_dropped = not_outliers.fillna(not_outliers.mean()).reset_index()

find_outliers_IQR(train['account_length'])

162    208.0
Name: account_length, dtype: float64

for feature in train:
    if train[feature].dtype != ('O'):
        plt.figure(figsize = (4,4))
        sns.boxplot(x ='churn', y = feature, data = train)
        plt.title(feature)
        plt.show()

1.4 Variaveis categoricas

# State vai ser um pouco mais complicado, pois temos 51 estados e fazer onehot encoding vai gerar muitos features...mas vou mesmo assim
# Vou fazer one-hot para os estados e para area_code
from sklearn.preprocessing import OneHotEncoder


onehot_area = OneHotEncoder()
onehot_area.fit(train[['area_code']])

# Train
encoded_values = onehot_area.transform(train[['area_code']])
train[onehot_area.categories_[0]] = encoded_values.toarray()
train = train.drop('area_code', axis=1)

# Test
encoded_values = onehot_area.transform(test[['area_code']])
test[onehot_area.categories_[0]] = encoded_values.toarray()
test = test.drop('area_code', axis=1)

# Lembrando que aqui TEM churn (Coluna EXTRA)
display(train.shape, train.head(2))

(4250, 19)

	index	state	account_length	international_plan	voice_mail_plan	number_vmail_messages	total_day_calls	total_day_charge	total_eve_calls	total_eve_charge	total_night_calls	total_night_charge	total_intl_calls	total_intl_charge	number_customer_service_calls	churn	area_code_408	area_code_415	area_code_510
0	0	OH	107.0	no	yes	26.0	123.0	27.47	103.0	16.62	103.0	11.45	3.0	3.70	1.0	no	0.0	1.0	0.0
1	1	NJ	137.0	no	no	0.0	114.0	41.38	110.0	10.30	104.0	7.32	5.0	3.29	0.0	no	0.0	1.0	0.0

display(test.shape, test.head(2))

(750, 18)

	id	state	account_length	international_plan	voice_mail_plan	number_vmail_messages	total_day_calls	total_day_charge	total_eve_calls	total_eve_charge	total_night_calls	total_night_charge	total_intl_calls	total_intl_charge	number_customer_service_calls	area_code_408	area_code_415	area_code_510
0	1	KS	128	no	yes	25	110	45.07	99	16.78	91	11.01	3	2.7	1	0.0	1.0	0.0
1	2	AL	118	yes	no	0	98	37.98	101	18.75	118	9.18	6	1.7	0	0.0	0.0	1.0

onehot_state = OneHotEncoder()
onehot_state.fit(train[['state']])

# Train
encoded_values = onehot_state.transform(train[['state']])
train[onehot_state.categories_[0]] = encoded_values.toarray()
train = train.drop('state', axis=1)

# Test
encoded_values = onehot_state.transform(test[['state']])
test[onehot_state.categories_[0]] = encoded_values.toarray()
test = test.drop('state', axis=1)

# Lembrando que aqui TEM churn (Coluna EXTRA)
display(train.shape, train.head(2))

(4250, 69)

	index	account_length	international_plan	voice_mail_plan	number_vmail_messages	total_day_calls	total_day_charge	total_eve_calls	total_eve_charge	total_night_calls	...	SD	TN	TX	UT	VA	VT	WA	WI	WV	WY
0	0	107.0	no	yes	26.0	123.0	27.47	103.0	16.62	103.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
1	1	137.0	no	no	0.0	114.0	41.38	110.0	10.30	104.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

2 rows × 69 columns

display(test.shape, test.head(2))

(750, 68)

	id	account_length	international_plan	voice_mail_plan	number_vmail_messages	total_day_calls	total_day_charge	total_eve_calls	total_eve_charge	total_night_calls	...	SD	TN	TX	UT	VA	VT	WA	WI	WV	WY
0	1	128	no	yes	25	110	45.07	99	16.78	91	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
1	2	118	yes	no	0	98	37.98	101	18.75	118	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

2 rows × 68 columns

# Vou trocar os que são apenas sim e não por 0 e 1 (0 = no, 1 = yes)
# Dataset de treino
train['international_plan'].replace(['no','yes'],[0,1],inplace = True)
train['voice_mail_plan'].replace(['no','yes'],[0,1],inplace=True)
train['churn'].replace(['no','yes'],[0,1],inplace = True)


# Dataset de teste
test.international_plan.replace(['no','yes'],[0,1],inplace = True)
test.voice_mail_plan.replace(['no','yes'],[0,1],inplace = True)

train.head()

	index	account_length	international_plan	voice_mail_plan	number_vmail_messages	total_day_calls	total_day_charge	total_eve_calls	total_eve_charge	total_night_calls	...	SD	TN	TX	UT	VA	VT	WA	WI	WV	WY
0	0	107.0	0	1	26.0	123.0	27.47	103.0	16.620000	103.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
1	1	137.0	0	0	0.0	114.0	41.38	110.0	10.300000	104.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
2	2	84.0	1	0	0.0	71.0	50.90	88.0	17.029889	89.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
3	3	75.0	1	0	0.0	113.0	28.34	122.0	12.610000	121.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
4	4	121.0	0	1	24.0	88.0	37.09	108.0	17.029889	118.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

5 rows × 69 columns

1.5 Variaveis numericas e ML

# Lembrando que CHURN esta bem desbalanceado (Olhar o grafico la em cima)
# Então ja vou resolver ele logo agora, depois dou scale no resto
from sklearn.model_selection import train_test_split


X = train.drop('churn', axis = 1).values
y = train['churn'].values

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state = 42)

Resolvendo o churn

from imblearn.over_sampling import SMOTE

sm = SMOTE(sampling_strategy = 1, random_state=1)
x_train_s, y_train_s = sm.fit_resample(X_train, y_train.ravel())

# creating the object of minmax scaler
from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)
test = scaler.transform(test)

c:\Users\Caio1\anaconda3\lib\site-packages\sklearn\base.py:413: UserWarning: X has feature names, but StandardScaler was fitted without feature names
  warnings.warn(

Random Forest

from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score
from sklearn.model_selection import GridSearchCV
from sklearn.metrics import confusion_matrix
from sklearn.metrics import classification_report


# Eu esqueci o parametros do RandomForest, so pra lembra
rf = RandomForestClassifier()
rf.get_params()

{'bootstrap': True,
 'ccp_alpha': 0.0,
 'class_weight': None,
 'criterion': 'gini',
 'max_depth': None,
 'max_features': 'sqrt',
 'max_leaf_nodes': None,
 'max_samples': None,
 'min_impurity_decrease': 0.0,
 'min_samples_leaf': 1,
 'min_samples_split': 2,
 'min_weight_fraction_leaf': 0.0,
 'n_estimators': 100,
 'n_jobs': None,
 'oob_score': False,
 'random_state': None,
 'verbose': 0,
 'warm_start': False}

grid = {'n_estimators': [200,300,400,500], 'max_depth': [2,5,10], 'min_samples_leaf': [0.1, 0.2]}
grid_rf = GridSearchCV(estimator = rf, param_grid = grid, cv = 5, n_jobs=-1)

# Fit 'rf' to the training set
grid_rf.fit(X_train, y_train)

# Predict the test set labels 'y_pred'
y_pred = grid_rf.predict(X_test)

display(accuracy_score(y_test, y_pred))
print(classification_report(y_test, y_pred))
display(confusion_matrix(y_test, y_pred))

0.8470588235294118

              precision    recall  f1-score   support

           0       0.85      1.00      0.92      1080
           1       0.00      0.00      0.00       195

    accuracy                           0.85      1275
   macro avg       0.42      0.50      0.46      1275
weighted avg       0.72      0.85      0.78      1275

c:\Users\Caio1\anaconda3\lib\site-packages\sklearn\metrics\_classification.py:1344: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\Users\Caio1\anaconda3\lib\site-packages\sklearn\metrics\_classification.py:1344: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\Users\Caio1\anaconda3\lib\site-packages\sklearn\metrics\_classification.py:1344: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))

array([[1080,    0],
       [ 195,    0]], dtype=int64)

grid_rf.best_estimator_

RandomForestClassifier(max_depth=2, min_samples_leaf=0.1, n_estimators=200)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

best_rf = grid_rf.best_estimator_

best_rf.fit(X_train, y_train)
y_pred = best_rf.predict(X_test)

accuracy_score(y_test, y_pred)

0.8470588235294118

Decision Tree

from sklearn.tree import DecisionTreeClassifier
dt = DecisionTreeClassifier()
dt.get_params()

{'ccp_alpha': 0.0,
 'class_weight': None,
 'criterion': 'gini',
 'max_depth': None,
 'max_features': None,
 'max_leaf_nodes': None,
 'min_impurity_decrease': 0.0,
 'min_samples_leaf': 1,
 'min_samples_split': 2,
 'min_weight_fraction_leaf': 0.0,
 'random_state': None,
 'splitter': 'best'}

grid = {'max_depth': [3, 4, 5, 6], 'min_samples_leaf': [0.04, 0.06, 0.08], 'max_features': [0.2, 0.4, 0.6, 0.8]}
grid_dt = GridSearchCV(estimator = dt, param_grid = grid,  scoring='accuracy', cv=10, n_jobs=-1)

grid_dt.fit(X_train, y_train)
y_pred = grid_dt.predict(X_test)
accuracy_score(y_test, y_pred)

0.8384313725490196

display(accuracy_score(y_test, y_pred))
print(classification_report(y_test, y_pred))
display(confusion_matrix(y_test, y_pred))

0.8384313725490196

              precision    recall  f1-score   support

           0       0.86      0.96      0.91      1080
           1       0.43      0.16      0.24       195

    accuracy                           0.84      1275
   macro avg       0.65      0.56      0.57      1275
weighted avg       0.80      0.84      0.81      1275

array([[1037,   43],
       [ 163,   32]], dtype=int64)

grid_dt.best_estimator_

DecisionTreeClassifier(max_depth=5, max_features=0.4, min_samples_leaf=0.04)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

XgBoost

len(y_train)

2975

np.unique(y_train, return_counts=True)[1][1]

403

import math
scale=round(math.sqrt(np.unique(y_train, return_counts=True)[1][0]/np.unique(y_train, return_counts=True)[1][1]),2)
scale

2.53

from xgboost import XGBClassifier

xgb = XGBClassifier(scale_pos_weight=scale, eval_metric ='mlogloss')
xgb.get_params()

{'objective': 'binary:logistic',
 'use_label_encoder': None,
 'base_score': None,
 'booster': None,
 'callbacks': None,
 'colsample_bylevel': None,
 'colsample_bynode': None,
 'colsample_bytree': None,
 'early_stopping_rounds': None,
 'enable_categorical': False,
 'eval_metric': 'mlogloss',
 'feature_types': None,
 'gamma': None,
 'gpu_id': None,
 'grow_policy': None,
 'importance_type': None,
 'interaction_constraints': None,
 'learning_rate': None,
 'max_bin': None,
 'max_cat_threshold': None,
 'max_cat_to_onehot': None,
 'max_delta_step': None,
 'max_depth': None,
 'max_leaves': None,
 'min_child_weight': None,
 'missing': nan,
 'monotone_constraints': None,
 'n_estimators': 100,
 'n_jobs': None,
 'num_parallel_tree': None,
 'predictor': None,
 'random_state': None,
 'reg_alpha': None,
 'reg_lambda': None,
 'sampling_method': None,
 'scale_pos_weight': 2.53,
 'subsample': None,
 'tree_method': None,
 'validate_parameters': None,
 'verbosity': None}

grid = {"learning_rate": [0.10,0.20,0.30 ], "max_depth":[ 3,5,10,20], "n_estimators" :[ 100, 200, 300, 500], 
              "colsample_bytree" :[ 0.3, 0.5, 0.7 ]}
grid_xgb = GridSearchCV(estimator = xgb, param_grid = grid, scoring = 'accuracy', return_train_score = True, verbose=1)

grid_xgb.fit(X_train, y_train)
y_pred = grid_xgb.predict(X_test)
accuracy_score(y_test, y_pred)

Fitting 5 folds for each of 144 candidates, totalling 720 fits

0.9482352941176471

display(accuracy_score(y_test, y_pred))
print(classification_report(y_test, y_pred))
display(confusion_matrix(y_test, y_pred))

0.9482352941176471

              precision    recall  f1-score   support

           0       0.95      0.99      0.97      1080
           1       0.92      0.73      0.81       195

    accuracy                           0.95      1275
   macro avg       0.93      0.86      0.89      1275
weighted avg       0.95      0.95      0.95      1275

array([[1067,   13],
       [  53,  142]], dtype=int64)

grid_xgb.best_estimator_

XGBClassifier(base_score=None, booster=None, callbacks=None,
              colsample_bylevel=None, colsample_bynode=None,
              colsample_bytree=0.7, early_stopping_rounds=None,
              enable_categorical=False, eval_metric='mlogloss',
              feature_types=None, gamma=None, gpu_id=None, grow_policy=None,
              importance_type=None, interaction_constraints=None,
              learning_rate=0.2, max_bin=None, max_cat_threshold=None,
              max_cat_to_onehot=None, max_delta_step=None, max_depth=10,
              max_leaves=None, min_child_weight=None, missing=nan,
              monotone_constraints=None, n_estimators=300, n_jobs=None,
              num_parallel_tree=None, predictor=None, random_state=None, ...)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

best_xgb = grid_xgb.best_estimator_

best_xgb.fit(X_train, y_train)
y_pred = best_xgb.predict(X_test)

accuracy_score(y_test, y_pred)

0.9482352941176471

1.6 Final

test.shape

(750, 68)

sampleS['churn'].replace(['no','yes'],[0,1],inplace = True)

y_pred2 = grid_xgb.predict(test)

y_pred2

array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0,
       0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0,
       0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1,
       0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0,
       0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 1, 1,
       0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0,
       0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1,
       0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0,
       0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0,
       0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0])

sampleS

	id	churn
0	1	1
1	2	0
2	3	0
3	4	1
4	5	1
...	...	...
745	746	1
746	747	1
747	748	0
748	749	1
749	750	1

750 rows × 2 columns

accuracy_score(y_pred2,sampleS['churn'])

0.6053333333333333